Leverage Unlabeled Data for Abstractive Speech Summarization with Self-Supervised Learning and Back-Summarization

TL;DR

This paper explores leveraging large amounts of unaligned data for French meeting abstractive summarization by using self-supervised pre-training and back-summarization techniques, significantly improving performance over baseline models.

Contribution

It introduces two novel approaches—self-supervised pre-training and back-summarization—for utilizing unaligned data in speech summarization, achieving substantial performance gains.

Findings

Large ROUGE score improvements over baseline

Effective use of unaligned reports for training

Combining methods yields best results

Abstract

Supervised approaches for Neural Abstractive Summarization require large annotated corpora that are costly to build. We present a French meeting summarization task where reports are predicted based on the automatic transcription of the meeting audio recordings. In order to build a corpus for this task, it is necessary to obtain the (automatic or manual) transcription of each meeting, and then to segment and align it with the corresponding manual report to produce training examples suitable for training. On the other hand, we have access to a very large amount of unaligned data, in particular reports without corresponding transcription. Reports are professionally written and well formatted making pre-processing straightforward. In this context, we study how to take advantage of this massive amount of unaligned data using two approaches (i) self-supervised pre-training using a target-side denoising encoder-decoder model; (ii) back-summarization i.e. reversing the summarization process by learning to predict the transcription given the report, in order to align single reports with generated transcription, and use this synthetic dataset for further training. We report large improvements compared to the previous baseline (trained on aligned data only) for both approaches on two evaluation sets. Moreover, combining the two gives even better results, outperforming the baseline by a large margin of +6 ROUGE-1 and ROUGE-L and +5 ROUGE-2 on two evaluation sets